Age-related cognitive decline is a problem of growing importance given the trend toward increased lifespan and the importance of cognitive function in determining risk for neurodegenerative disease. Despite the importance of this problem, the cognitive and molecular changes that mediate it are still poorly understood. While there are competing theories for the mediators of cognitive aging at both psychological and molecular levels, theories on aging and the research supporting them have typically focused on a single level of analysis. As our understanding of the biological basis for behavior grows, this divide becomes less sensible. Instead, psychological theory should be constrained according to its known biology, and biology should in turn inform psychological theory. Here I propose to learn neural network modeling and magnetic resonance spectroscopy (MRS) techniques in order to bridge human psychological theory that has been the focus of my recent work to the molecular level theory that was the focus of my post-baccalaureate training at the NIA. I will build from my recent work that highlights a key statistical problem faced by the brain: how to selectively pool information across relevant sources but partition information across irrelevant ones. I will closely examine the cognitive and molecular mechanisms that allow for efficient pooling and partitioning to test the overarching theory that, due to impaired glutamate and dopamine signaling, older adults develop a selective deficit in pooling relevant sources of information. I will test this theory in two separate paradigms: visual working memory (K99) and learning and perceptual inference (R00). During the K99 phase of the award I will examine how pooling information from visual targets with similar features can 1) improve effective memory capacity, 2) be achieved by a neural network model, and 3) be impaired by simulated molecular deficiencies (glutamate, dopamine, norepinephrine). This bridge between cognitive and molecular levels of analysis will be used to test whether age-related memory deficits are due to inefficient pooling and mediated by molecular deficits in glutamate and dopamine (as measured through MRS and behavioral proxy, respectively). During the R00 phase of the award I will use the training in neural networks and MRS provided in the K99 phase to examine how 1) pooling sequential pieces of information affects learning and perceptual bias, 2) efficient pooling and partitioning can be achieved by a neural network model and 3) these processes are disrupted by specific molecular deficiencies. The established relationships between statistical properties (sequential pooling and partitioning), psychological measurements (learning and perceptual bias) and molecular factors (glutamate, dopamine, and norepinephrine signaling levels) will be used to test whether age-related differences in learning and perceptual bias reflect deficient pooling mediated by local glutamate deficiency (as measured through MRS). Overall, this work will contribute a deeper and more detailed understanding of the psychological and molecular factors that interact to mediate age-related cognitive decline. In addition, the K99 training will provide me the tools necessary to link psychological and molecular levels of analysis and the R00 phase will build the foundations of my independent research laboratory allowing me to develop a successful scientific career driven by experiments that build and refine a unified understanding of cognitive aging.